Journal / Research note · 7 min read
Why we rebalance weekly: daily, weekly, monthly and quarterly, tested
How often should a rules-based ETF portfolio rebalance? Our tests of daily, weekly, monthly and quarterly schedules, with costs, drawbacks and exceptions.
By Lowdraft. Published October 20, 2026.
Every Keel has to answer a plain question: how often should it look at the market and trade? Check too rarely and a crash arrives before the rules can react. Check too often and you pay for noise. We tested four schedules on our main Keels, counted trading costs, and checked the answer on every weekday. This is what we found, including the parts that do not flatter weekly.
How we tested
- Period: August 2019 to October 2026, using the longest history all the holdings share.
- Execution: a signal on one day trades the next day, so the test never trades on information it could not have had.
- Cost: 5 basis points (0.05%) of the amount traded, as a base case. We also tried 2 and 10.
- Yardsticks: UPI first (return per unit of how deep and how long the drawdowns were), then maximum drawdown and annual return (CAGR).
- Weekly is averaged over all five weekdays. A Keel that trades on Mondays and one that trades on Wednesdays can differ by a lot, so a Monday-only result flatters the schedule.
The result for our lead Core + Boost Keel
| Schedule | Rebalances a year | UPI | Annual return | Worst drawdown |
|---|---|---|---|---|
| Daily | 252 | 8.5 | 19.5% | −11.3% |
| Weekly (average of 5 weekdays) | 53 | 8.7 | 20.3% | −11.5% |
| Monthly (average of month-end and month-start) | 12 | 5.3 | 18.9% | −12.9% |
| Quarterly | 4 | 1.3 | 12.0% | −28.5% |
Hypothetical back-test, 5 bp per trade. Not a forecast.
Why not daily
Before any costs, daily and weekly were about even. Daily was a touch better at the lowest cost we tried and a touch worse at the highest. Daily did not give us a shallower drawdown either (−11.3% against −11.5%). What it did give us was about five times as many rebalances, and a result that gets worse as costs rise.
| UPI by cost per trade | 2 bp | 5 bp | 10 bp |
|---|---|---|---|
| Core + Boost, daily | 9.4 | 8.5 | 7.2 |
| Core + Boost, weekly | 9.0 | 8.7 | 8.1 |
| Core + Boost, monthly | 5.6 | 5.3 | 5.0 |
| Core + Boost, quarterly | 1.4 | 1.3 | 1.3 |
| Income, daily | 4.4 | 3.7 | 2.6 |
| Income, weekly | 4.1 | 3.5 | 2.7 |
| Income, monthly | 4.2 | 3.9 | 3.5 |
| Income, quarterly | 0.4 | 0.4 | 0.4 |
Our Keels trade large ETFs, which usually have tight spreads, but real fills include slippage and delay. The more you trade, the more of the tested result you hand back. A weekly schedule has about the same protection as daily at a fraction of the trading, so it leaves more room for reality to be unkind.
Why not monthly
- It reacts too late. Our crash switch looks at a 10-day fall. Checking once a month means it can be days or weeks old when the Keel finally acts.
- The result depends on the calendar. The same Keel rebalancing at month-end scored UPI 4.4, and at month-start 6.3. A schedule whose outcome depends on which day of the month you pick is partly luck.
- It gives up the gain from trend changes. Annual return dropped about a point and a half and the drawdown deepened, for a modest saving in trading.
Why not quarterly
Quarterly was clearly the worst. With only four decisions a year, the Keel could not step aside in time for the February to March 2020 crash, and its worst drawdown was about −28.5%, against about −11.5% for weekly. Returns fell to about 12% a year. A rule written to avoid deep drawdowns needs to run often enough to act on them.
Where weekly is not the answer
- Slower, income-style Keels. For our Income Keel, monthly matched weekly at low cost and beat it at higher cost (UPI 3.5 against 2.7 at 10 bp). These Keels hold calmer funds and change their minds less often, so extra checking adds trades, not insight.
- High-cost holdings. In our tests weekly kept its return edge over monthly until costs reached roughly 15 bp per trade. Thinly traded ETFs can cost more than that.
- Taxable accounts. More rebalances can mean more taxable events. Retirement accounts avoid that issue, which is one reason we favor ETFs and IRAs for this kind of rule.
The weekday problem, and what we do about it
Within weekly, the day you trade matters more than the gap between daily and weekly. For the Core + Boost Keel, the five weekdays produced UPI between 7.6 and 10.0 at the same cost. Monday and Tuesday looked best and Wednesday worst, which is partly luck. We therefore treat the five-day average as the honest expectation, and plan to split a real portfolio across weekdays so no single day's luck decides the outcome.
What we take from it
- Weekly is a good default for trend-and-crash Keels: frequent enough to act in time, infrequent enough to keep costs low.
- Daily is not worth it unless trading is nearly free, and even then it did not reduce drawdowns.
- Monthly can suit slower, income-style Keels. Test it for yours.
- Avoid quarterly for any rule meant to limit drawdowns.
- Always test your costs. The right schedule depends on what a trade costs you.
Open this note as its own page
Hypothetical results from our research, prepared with hindsight and partly fitted to past data. Costs, slippage, delays and taxes in real trading reduce results. Not investment advice. Past performance, real or simulated, does not predict future results. You can lose money.